Direct-search methods for decentralized blackbox optimization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bergou, El Houcine, Diouane, Youssef, Kungurtsev, Vyacheslav, Royer, Clément W.
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915895759077376
author Bergou, El Houcine
Diouane, Youssef
Kungurtsev, Vyacheslav
Royer, Clément W.
author_facet Bergou, El Houcine
Diouane, Youssef
Kungurtsev, Vyacheslav
Royer, Clément W.
contents Derivative-free optimization algorithms are particularly useful for tackling blackbox optimization problems where the objective function arises from complex and expensive procedures that preclude the use of classical gradient-based methods. In contemporary decentralized environments, such functions are defined locally on different computational nodes due to technical or privacy constraints, introducing additional challenges within the optimization process. In this paper, we adapt direct-search methods, a classical technique in derivative-free optimization, to the decentralized setting. In contrast with zeroth-order algorithms, our algorithms rely on positive spanning sets to define suitable search directions while still possessing global convergence guaranties, thanks to carefully chosen stepsizes. Numerical experiments highlight the advantages of direct-search techniques over gradient-approximation-based strategies.
format Preprint
id arxiv_https___arxiv_org_abs_2504_04269
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Direct-search methods for decentralized blackbox optimization
Bergou, El Houcine
Diouane, Youssef
Kungurtsev, Vyacheslav
Royer, Clément W.
Optimization and Control
90C56
Derivative-free optimization algorithms are particularly useful for tackling blackbox optimization problems where the objective function arises from complex and expensive procedures that preclude the use of classical gradient-based methods. In contemporary decentralized environments, such functions are defined locally on different computational nodes due to technical or privacy constraints, introducing additional challenges within the optimization process. In this paper, we adapt direct-search methods, a classical technique in derivative-free optimization, to the decentralized setting. In contrast with zeroth-order algorithms, our algorithms rely on positive spanning sets to define suitable search directions while still possessing global convergence guaranties, thanks to carefully chosen stepsizes. Numerical experiments highlight the advantages of direct-search techniques over gradient-approximation-based strategies.
title Direct-search methods for decentralized blackbox optimization
topic Optimization and Control
90C56
url https://arxiv.org/abs/2504.04269